I am running a morphometry study on a public dataset with most, but not all, subjects having a T1 and T2 (about 75% have both, remaining only have T1). To get the best surfaces possible, I have been including the T2 in recon-all process when able to. However, this means subjects with only a T1 have underwent a different processing workflow. ICV is an important predictor in my study, and I was wondering if the presence of the T2 image in recon-all should impact the ICV estimates? As I see it, I have the following options:
- T2 is a factor, so I use only include T1 in recon-all so all subjects have
the same pipeline
- T2 is a factor, so I harmonize the data as I would with data from different
- T2 is a factor, so I only analyze subjects with T1 + T2
- T2 is not a factor, just use all subjects with no special considerations
I am also running vertex-wise surface based morphometry, where I imagine pial refinements could be more impactful. I suppose in that case, I should harmonize the morphometry data, but was curious if something like that is necessary.